Optimization And Empirical Open-end Fund Value At Risk Measurement Models

Our country’s open-end fund requires higher risk management. VaR is the main indexto measure current financial risk. The calculation has a lot of kinds of methods. The Copulamodel can depict the variables of nonlinear, non related with symmetrical structure, anddoes not limit the edge distribution types, which measure the risk value model based on thetheory of Copula value measurement of portfolio risk compared to the traditionalmultivariate distribution model to measure more accurately, according to the existingGARCH-Copula model, through the introduction of Gauss kernel function estimation andGPD model to improve the accuracy, the fat tail distribution fitting multivariate marginaldistribution, and the leverage effect of volatility clustering, builds the optimization modelof TARCH-GPD-Copula. This paper presents an optimization model to retain the originalmodel which can describe the marginal distribution of volatility clustering advantages, atthe same time, make up the original model can not fully describe the defect distributionpeak thick tail edge, improves the fitting accuracy of multivariate joint distribution, whichcan more accurately measure portfolio assets value of VaR. Optimization model oftheoretical significance lies in the universal calculation, applicable to all types of financialassets investment portfolio value at risk.This paper firstly put forward the necessity of using measure theory of value at riskCopula based on the current situation of China’s open-end fund risk management theory,introducing Copula theory and extreme value theory. Secondly, in the framework offinancial risk measure, in-depth analysis of the advantages and disadvantages ofGARCH-Copula model, TARCH-GPD-Copula model based on the measure of VaR, andgives the solving steps and the source program.The empirical research in this paper, choose a large selection of Chinese Hybridopen-end fund heavily invested ten stock portfolio as the object of study, usingTARCH-GPD-Copula model, calculated the VaR values by Monte Carlo simulationmethod. At the same time, GARCH-Copula model, variance covariance method, historicalsimulation method of portfolio selection under VaR is calculated. Compare the fourmodels’validity of calculated VaR values through the of Kupiec test. Results show that theoptimal model of TARCH-GPD-Copula posterior testing effect is best, the closest to thetrue value of risk estimation of value at risk, lowest failure rate. The empirical results areconsistent with the theory, the theory is correct that proved from the empirical point ofview，hope for it can provide the reference for the open-end fund investment portfolio risk measure for our fund management company or the fund supervision department.